Background <p>Clinical serum biochemistry is used to assess maternal and fetal health; however, it involves complex physiological adaptations during fetal development, challenging obstetricians. This study aimed to understand these physiological variations during pregnancy and evaluate the need and feasibility of establishing appropriate reference intervals (RIs).</p> Methods <p>Thirty-four serum biochemical indicators related to renal, liver, cardiac, glycolipid metabolism, and electrolyte systems in 4390 healthy pregnant women from Fujian, China were analyzed. Data were obtained from the obstetric big-data platform of Fujian Maternal and Child Health Hospital in China. After preliminary judgement by visualizing gestational trends in a scatter chart, the gestational- or maternal-age divisions for each indicator were determined in combination with classification tree algorithms and the Harris–Boyd criterion. New RIs corresponding to the newly determined divisions underwent real-world validation and were compared with traditional trimester-based RIs by a bootstrap-based independent sample test.</p> Results <p>Each indicator exhibited a dynamic trend throughout pregnancy, revealing discrepancies with traditional trimesters. Based on re-determined divisions, new RIs were established and validated using real-world data. Significant differences were identified between the traditional and new RIs in urea, creatinine, uric acid, alkaline phosphatase, glucose, total cholesterol, and apolipoprotein A in gestational-age divisions and those in albumin in maternal-age divisions (all <i>P</i> &lt; 0.001).</p> Conclusions <p>Our findings suggest that identifying physiological fluctuations based on actual biochemical indicators may be more appropriate than relying on traditional trimester-based classification standards when establishing pregnancy-specific RIs.</p> Trial registration <p>Not applicable.</p>